ERSA Congress 51st European Congress of the Regional Science Association 30th August - 3rd September 2011 Barcelona, Spain The role of space in urban housing market João Lourenço Marques E-mail: jjmarques@ua.pt Eduardo Anselmo de Castro Department of Social, Juridical and Political Sciences (Research Unit of Governance, Competitiveness and Public Policy University of Aveiro, Portugal Arnab Bhattacharjee Department of Economic Studies University of Dundee, UK ABSTRACT This paper emphasizes the quantitative analysis of space in relation to hedonic housing price models. Three aspects of space will be highlighted: i) spatial heterogeneity (spatial patterns): hedonic housing amenities may be valued differently in different locations which are related to specific housing sub-markets; ii) spatial dependence (spillovers): the degree by which price increases (or decreases) in a given sub-market is influenced by other sub-markets, or by another property within the same sub-market. iii) spatial scale: the study of heterogeneity and spillovers crucially depends on the level of geographical scale at which submarkets are defined. In the literature the difficulty of defining sub-markets and understanding the relationship between them is broadly identified, and appropriate methods for defining housing markets are also presented. However, there is not a consensus on which methodologies should be used. As a contribution to understand spatial structure (heterogeneity and spillovers) in urban spaces some empirical results will be presented. A new methodology to analyse spatial spillovers [rather than an ex ante definition of a spatial weight matrix (W)] will be developed. This procedure based on non-parametric approach will be applied to a rich database. An interesting outcome of this methodology is possibility of finding meaningful values of negative interaction. 1.INTRODUTION This paper aims to analyze the importance of space in the housing market using a spatial hedonic pricing model applied to the urban and sub-urban areas of Aveiro. Spatial interactions (spillovers effects), spatial heterogeneity and spatial scale are important aspects to analyse housing market. The first two aspects (heterogeneity and dependence) are widely stressed in the spatial econometrics literature; however, the common practice of representing the spatial interactions, using a weight matrix (W) a priori defined [1-8], has been often expressed as inadequate. The traditional approach to characterize spatial interactions is to define a matrix W, which represents theoretically and ex ante defined forces of interdependence (spatial autocorrelation), usually modelled by functions of distance or contiguity. Since the spatial interactions may be driven by other intangible factors (economic,